Samples file¶

The samples file names each sample and where its data is. Its extension picks the format: .tsv (or anything not .yml/.yaml) for the two-column table, .yml for the structured form.

TSV¶

No header. Two or three whitespace-separated columns:

# sample_id   run_path                                   [reference]
wt_rep1       /beevol/data/nanopore/20260901_gDNA_wt
wt_rep1       /beevol/data/nanopore/20260908_gDNA_wt_topup
mut_rep1      /beevol/data/nanopore/20260901_gDNA_mut    /beevol/data/ref/other.fa

run_path is a run directory (POD5 found under pod5_pass, pod5_fail, pod5, pod5_skip, recursively), a directory holding .pod5 files directly, or a single .pod5 file.
A sample id on several lines has several inputs; they are basecalled together.
The optional third column overrides reference.fasta for that sample.

YAML¶

runs:
  # Pooled run the pipeline basecalls: one dorado pass with --kit-name, then demux.
  - path: /beevol/data/nanopore/20260901_pooled
    kit: SQK-NBD114-24
    reference: /beevol/data/ref/chm13v2.0.fa          # optional: for every sample of this run
    samples:
      wt_rep1: barcode01
      mut_rep1: {barcode: barcode02, reference: /beevol/data/ref/other.fa}

  # MinKNOW live-basecalled run: bam_pass/<barcode>/*.bam used as-is.
  - path: /beevol/data/nanopore/20260910_pooled_live/results
    basecalled: true
    samples:
      fork_12: barcode12
      fork_15: barcode15

  # Unbarcoded run: the whole run is one sample.
  - path: /beevol/data/nanopore/20260815_gDNA
    samples:
      control: ~

Key	Meaning
`path`	run directory (as for TSV)
`kit`	dorado `--kit-name`; required when the run's samples carry barcodes and the pipeline basecalls
`basecalled`	`true`: skip dorado, collect MinKNOW's `bam_pass` (and `bam_fail` with `minknow.include_fail`)
`reference`	per-run reference override; a sample can override again
`samples`	`name: barcode`, `name: {barcode, reference}`, or `name: ~` for the whole run

Modes¶

Every sample resolves to exactly one mode, printed at startup (Samples: wt_rep1 [basecall], ...):

Mode	Input	Rules
`basecall`	POD5 per sample	`link_sample_pod5` → `basecall`
`demux`	POD5 per barcoded run	`link_run_pod5` → `basecall_run` → `demux_run` → `collect_demuxed_sample`
`prebasecalled`	MinKNOW BAMs	`collect_minknow_bams` (POD5 is still linked, for DNAscent)

A sample cannot mix modes (one input barcoded, another not); split it into two samples. A prebasecalled sample with a null barcode on a barcoded run takes only the BAMs directly under bam_pass/, never the per-barcode subdirectories, so barcodes are never pooled by accident.

Sample ids must match [A-Za-z0-9][A-Za-z0-9._-]*; they become file and directory names.